Changes between Initial Version and Version 1 of Ticket #23139, comment 70


Ignore:
Timestamp:
08/18/17 19:00:41 (2 years ago)
Author:
zgershkoff
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #23139, comment 70

    initial v1  
    33Yes, but in two different ways. Some tests fail because the ground set labels are different, and some tests fail because it wants a `BasisExchangeMatroid` and gets a `GraphicMatroid`.
    44
    5 As I understand it, then the constructor takes a graph to make a `RegularMatroid`, it will use vertex tuples as ground set labels unless there are multiedges, where it will use integers instead. I think this is the right behavior for representing a graph with a matrix, since we lose the context of the graph, but I think it's unnecessary when we still have the graph attached to the matroid.
     5As I understand it, when the constructor takes a graph to make a `RegularMatroid`, it will use vertex tuples as ground set labels unless there are multiedges, where it will use integers instead. I think this is the right behavior for representing a graph with a matrix, since we lose the context of the graph, but I think it's unnecessary when we still have the graph attached to the matroid.
    66
    77The method `groundset_list()` belongs to `BasisExchangeMatroid`, which is a superclass of `RegularMatroid`. It looks like other matroid classes don't have it, and it doesn't seem like it would translate easily since only `BasisExchangeMatroid`s have the attribute `_E`. When orlik-solomon.py was written, they were probably expecting a regular matroid, which is why I changed the test to give it one.